BIG DATA & GENDER BRIEF SERIES

Street Harassment and Women’s Educational Choices: A Geospatial Analysis GIRIJA BORKER, WORLD BANK/BROWN UNIVERSITY

IntroductionHow does the threat of street harassment affect

women’s lives? Using a combination of student

surveys, Google Maps data, and crowdsourced

information from mobile applications, this

study looks at how safety concerns influence

educational choices among women in New Delhi,

India. The research finds that women choose

poorer quality colleges, spend considerably

more on transportation, and accept longer

commute times in order to travel by routes

that are perceived to be safer. In addition to

the risks of harassment and assault, an unsafe

public sphere inflicts serious educational and

economic consequences on women. These costs

are felt both immediately and over the course

of a lifetime, in the form of reduced labor force

participation and earnings.

Our ApproachDelhi University (DU) is an umbrella entity

comprised of several colleges spread across

the city. Each college has its own campus and

operates like an independent university, and

selectivity in admissions provides a reliable

indicator of college quality. A survey of 4,000 DU

students identified the “choice set” of colleges

available for each student, or the set of colleges

that a student is eligible to attend based on their

high school exam scores. The colleges captured

in the sample span the range of quality in DU

as a whole and the students in the sample are

representative of the wider student body in the

University.

An algorithm developed for this study used

Google Maps to map all possible routes available

for students to take to each college in their choice

set, where routes are defined as a combination

of landmarks and travel modes.

Figure 1. Safety surface of the Delhi metropolitan area.

These routes are overlain on a “safety surface”

(Figure 1) constructed using a crowd-sourcing

mobile application, SafetiPin. Over 26,500

SafetiPin audits from November 2013 to January

2016 provided information about openness of

spaces, visibility, presence of security personnel,

condition of the walking path, presence of other

people (especially women and children), access

to public transport, lighting, and the overall

feeling of safety. Students’ actual and potential

routes are assigned a safety score on the basis

of the safety surface and safety associated with

a travel mode. The latter is estimated using data

from Safecity, which has over 5,500 reports of

harassment on different modes of transport,

including buses, rickshaws, metro, and walking.

A comparison of male and female students with

similar characteristics (same neighborhood, major,

admission year, etc.) allows comparative analysis

of the costs students pay — in terms of quality of

college chosen, money, and time — for safety.

SafetyLow

Medium

High Kilometers

0 2 4 8 12 16

ResultsDespite having better scores on high school

exams, women choose worse quality colleges than

men — both in absolute terms and within the set

of colleges for which they are eligible. Generally,

women are willing to trade a college in the top

20% of quality for one in the bottom 50% — on

average, nearly nine ranks lower in their choice

set — in order to travel by a route that offers about

a one standard deviation (SD) increase in safety,

which represents around a 3.1% decrease in the

risk of rape. Men, on the other hand, choose

colleges only about one rank lower in their choice

set for the same degree of added safety. Figure 2

illustrates these gendered differences in decision

making.

Figure 2. Gendered educational choice.

The horizontal axis represents differences in high

school exam scores between students and their

neighbors (other students of same gender, from

the same residential area, studying the same major

with the same admission year).

A higher score gap bin represents higher-achieving

students with an expanded choice set of colleges

who, absent safety concerns, should choose

higher-quality colleges. The vertical axis shows

differences in the safety of the chosen travel route

(Panel A) and quality of attended colleges (Panel

B) between students and neighbors. The graph in

Panel A shows that as their choice set expands,

female students (red dotted line) choose safer

routes to travel by compared to males (solid blue

line). Panel B shows the relationship between

high school quality and college choice is weak for

women, and much stronger for men.

Women also spend INR 20,000 (USD 310) more

per year for added safety, nearly 16 times as much

as men — a gap that is nearly double the average

annual tuition at DU, and about 75% of annual per

capita income in Delhi. Women are also willing to

travel 40 minutes more daily for a safer route, in

comparison to just four minutes more for men.

ImplicationsIn Delhi, 95 percent of women between the ages

of 16 and 49 report feeling unsafe in public spaces.

In the sample used in this study, nearly nine out

of ten female college students have faced some

form of harassment, and 40% have been touched,

groped, or grabbed. Public areas represent serious

daily risks to the security of women and girls.

This study shows that the indirect costs of

harassment are also very high. Choosing a worse

quality college affects later academic training, peer

network, access to jobs, and lifetime earnings —

and may affect aggregate economic productivity

for a society. In India, labor force participation

rates for working-age women have stagnated at

26-28% in urban areas between 1987 and 2011.

The drivers of this low participation rate are not

clear, but lack of physical security in public areas

may be a contributor. This study finds that policies

to increase safety of travel routes — improving

street lighting, funding self-defense programs, and

assuring security on public transit, for example —

can have powerful impacts on reducing gender

gaps in school quality, costs of transport, and

commute time.

BIG DATA & GENDER BRIEF SERIES

20

0

0 5 10Score gap bin

15

10

15

5

Qu

alit

y d

i�e

ren

ce

B

1

-1

0 5 10Score gap bin

15

0

.5

-.5Saf

ety

di�

ere

nce

A

Females Males